building an Agentic AI platform for Russia's largest retailer**|Company TopSelection

We are looking for an engineer who designs multi-agent systems, is familiar with RAG, evals, and LLM security, not just someone who calls the OpenAI API. What we do: A unified AI ecosystem within the company: corporate RAG search, multimodal generation, autonomous agents, integration with ERP/CRM, access control, and real-time model auditing. Tech Stack: Transformers, RAG, Agentic Workflows, LangChain/LangGraph, vector search, tool calling, structured output, evals, security (prompt injection, PII), LoRA/PEFT, GPU inference, Airflow, Kafka, K8s, MLflow, Prometheus, Grafana.

Conditions: 💸** 240,000 – 260,000 RUB Gross** 🕒 Fulltime 🌍 Remote, strictly Russian Federation, located in Russia ⏰ Work according to Moscow time 📃**Employment: **via Individual Entrepreneur contract

Team: Project from scratch, greenfield, full spectrum of tasks — from architecture to production monitoring. Responsibilities: — Design the core of the AI platform (agent orchestration) — Develop SDK/tooling, agent skills registry service — Build Knowledge Lake API + connectors to Confluence, SharePoint, Yandex Wiki — Optimize economics, latency, scalability Requirements: ✔️ Transformer architecture ✔️ RAG + embeddings + vector search ✔️ Agentic / multi-agent systems (state, context, human-in-the-loop) ✔️ Evals, hallucination control, drift monitoring ✔️ LLM Security (prompt injection, PII, access auditing) ✔️ OpenAI / Anthropic API, Hugging Face ✔️ Highload systems (e-commerce / retail — welcome) ✔️ Airflow, MLflow, Kafka, Redis, K8s, Prometheus, Grafana ✔️ GPU inference optimization, LoRA/PEFT ✔️ Experience with personal data and privacy